I have a simple python script that buffers a parcel and returns a CSV of parcels within a specified distance of the original parcel. This is published as a service and seems to work very well. My problem is that when I publish this script as a service, the publishing tool makes a complete (unnecessary) copy of my parcel geodatabase. Then continues to use the copied file geodatabase rather than the registered geodatabase. The script was published in March 2014, so the data is getting somewhat stale. I do have the parcel geodatabase registered in the data store, but when published it does not recognize it.
How do I publish a script to use the geodatabase and not make a copy? I would appreciate any help!